Quarterback Risk vs. Reward

2018 Review and 2019 Outlook

It took me a lot longer to get around to this than I had planned. Having my first “big boy job” decreased my desire to sit at my computer for a few days and copy and paste NFL game logs into an Excel spreadsheet; nevertheless, said “big boy job” has given me an opportunity to sharpen my spreadsheet and programming skills to get to the point that I could rig up my workbook so that entering in the 2018 Season’s statistics, which used to take me a few days, only took a couple hours.

So here I am, back on the horse – ready to analyze, speculate, and write again.

For those that are unfamiliar with my studies, I have created a metric based entirely off box scores that I call the Quarterback Productivity Rating (QPR or productivity rating, for short). This rating is set to combine the simplicity and lack of bias of the passer rating with better encompassment of the NFL quarterbacks’ duties, as ESPN’s mysterious Quarterback Rating, does.

Statistics cannot grade a quarterback better than film study can, nor can box score analysis; however, for those of us limited on the resources of time and knowledge required to study every quarterback in the NFL, this rating can give us a better idea of how quarterbacks compare from a productivity standpoint and why. To learn more about the QPR, including how it is calculated, and my studies of quarterbacks, please read my MTSU Honors Thesis, The Worth of NFL Free Agent Quarterbacks.

To outline and reference for the rest of this article, I have calculated every starting quarterback’s productivity rating since 2011, so I will refer to “since 2011” plenty for that reason. I have picked 2011 as the start because that was the beginning of the new CBA and new rookie contract scaling. I will also refer to “qualifying starters” as quarterbacks that started at least four games in a season.

To start our “since 2011”s, the average QPR per start in 2018 was a 32.16, which is the highest since 2011, minus 2015, which was an ever-so-slightly-and-not-statistically-significantly-higher 32.22. Compare 2018’s 32.16 mean to 2017’s mean of 27.80, we can almost assuredly say that quarterbacks were more productive on average this year.

The Rookies and the Elites

I want to continue off of what I learned from my thesis (and honestly what we are all now seeing with the naked eye), and that is that NFL teams need to try to win while their quarterbacks are on rookie contracts. The “Super Bowl Window” is no exaggeration or myth.

Of the qualifying starting quarterbacks with the top 12 win percentages in 2018:

• six came from teams with Day One starters on rookie contracts (Jared Goff, Nick Foles, Mitch Trubisky, Patrick Mahomes, Deshaun Watson, and Dak Prescott)
• five are considered elite quarterbacks (Drew Brees, Philip Rivers, Tom Brady, Andrew Luck, and Russell Wilson)
• the last is Lamar Jackson

Jackson is the outlier here. While he is on a rookie contract, the Baltimore Ravens’ Day One starter of Joe Flacco took up over 14% of the team’s cap space in 2018, per Spotrac, which is well over the 6.14% average in 2017. Foles does not count as an outlier, as his Philadelphia Eagles’ Day One starter of Carson Wentz is on a rookie deal, and Foles was on an inexpensive deal, as well.

As I discussed in my thesis, if a team is not going to start a quarterback on a rookie deal, then they better be sure that their starter is elite. NFL teams have realized that they need to either have an elite quarterback or surround a passer on a rookie contract with the best supporting cast possible.

To Sign or Not To Sign - Russell Wilson

The Seattle Seahawks are facing a very difficult situation. Franchise QB Russell Wilson has present them with a mid-April deadline to complete contract negotiations. Does Seattle make Wilson the highest paid quarterback, or do they move on?

Since 2011, Wilson is the third-winningest starter with at least 16 games under his belt, winning over 67% of his starts, which is ridiculous; however, his productivity rating is not as superb. His 35.60 average ranks as the thirteenth-highest of those starters, in-between the overpaid Kirk Cousins and Matthew Stafford.

What the box score does not show, however, is that Wilson's QPR is superb given the offensive line and running backs that he had prior to 2018. Sure, the defense gave him opportunities, but Wilson put the offense on his back. The QPR rating does not show that Wilson had his least pass attempts since 2013, and his 65.6% completion percentage was the second-highest of his career. So, while the QPR analysis does show that he is a Top 15 quarterback since the new CBA, it does not show that he is elite and deserves a new contract.

Predicting Sophomore Leaps and Slumps

Trying to predict how a quarterback’s statistics will improve is very difficult, especially because of how many factors contribute to those statistics; however, we can still try to do so.

Looking at all 31 quarterbacks that started at least one game in each of their rookie and sophomore seasons, we can see that not one rookie that posted a starting mean QPR above 30.00 saw an improved QPR in their second season, and of those seven, only Russell Wilson, Cam Newton, and Dak Prescott produced mean QPRs above 30 in Year 3.

So, it’s not absurd to think that the class’s lone outstanding rookie, Baker Mayfield, will see some regression. That said, Mayfield and the Cleveland Browns could break the mold by adding an outstanding piece in Odell Beckham Jr. Only five quarterbacks drafted since 2011 saw increased QPRs in each of their first three seasons, and only Ryan Tannehill of this bunch is not a starter in the NFL today.

Looking back this year’s rookies, only Mayfield produced a mean QPR above 30, while the rest were all below 23.00, which would be on average a touchdown less per start. Per the model I created using the aforementioned 31, Mayfield is due for regression, and the others are due for major increases in productivity; however, while Mayfield may be due for regression, he is projected to win nearly 55% of his starts, which could lead the Browns to the playoffs for the first time since 2002.

Our win percentage projection model for second-year quarterbacks has a correlation coefficient of 0.7595, which we want to see as close to 1.00 as possible. This is a strong correlation coefficient, though, and we can see strong relation between second-year quarterbacks’ mean productivity and win percentage. This correlation is much stronger than the correlation coefficient between all starting quarterbacks and their win percentages, which sits at 0.4587. So, if Baker can produce similarly to how he did as a rookie, the Browns can expect success.

The Not-So-Elite

Failing to follow the elite-or-cheap strategy can look two different ways: paying a non-elite quarterback big money or starting a quarterback on a rookie deal and not surrounding them with the pieces necessary for success.

Of the qualifying starting quarterbacks with the bottom 13 win percentages in 2018, there were:

• three quarterbacks from teams where the Day One starter was on a rookie contract (Sam Darnold, Jameis Winston, and Ryan Fitzpatrick)
• four starters on large contracts despite not being elite (Derek Carr, Blake Bortles, Eli Manning, and Matthew Stafford)
• three backups for non-elite quarterbacks on non-rookie contracts (C.J. Beathard, Nick Mullens, and Jeff Driskel)
• one elite quarterback – Aaron Rodgers

We have a few more outliers here than with the Top 10. Rodgers is obviously an outlier, being a Hall of Fame quarterback; however, it is arguable that the Green Bay Packers have not done a good job of surrounding him with talent to win games or pairing him with good coaching. It is hard to fault the 49ers for Beathard’s and Mullens’s low win percentages, as their starter in Jimmy Garoppolo has shown elite potential and won all of his games that he had started prior to the 2018 season.

The others all fit the mold – the New York Jets and Tampa Bay Buccaneers did not surround Darnold and Winston, respectively, with the pieces necessary to win games, and ultimately, their head coaches were fired this offseason. Carr, Bortles, Manning, and Stafford are all not elite, yet their teams hitched their wagons to them; likewise, the Cincinnati Bengals committed to the lackluster Andy Dalton, whose injury only set them up for further failure, despite his winning only five of the 11 games that he started.

The Big QB Move

Who is set up for the aforementioned type of failure in 2019? The Jacksonville Jaguars, of course. In eight Regular Season starts since 2017, Foles has averaged a mediocre QPR of 28.84 with a wild standard deviation of 31.79. His 75% winning percentage can be attributed to the superb team that the Eagles built around rookie starter Wentz. The Jaguars have some good supporting pieces, especially on the defensive side… but is it enough? The talk following the team’s AFC Championship loss to the Patriots on January 21^st, 2018, was that they were just a quarterback away from becoming a Super Bowl caliber team.

Rather than drafting a quarterback, they double downed on Blake Bortles, awarding him a 3-year, $54M contract, rather than drafting a quarterback early in the 2018 NFL Draft, where they could have drafted Heisman winner Jackson at the 29^th spot in the Draft. I don’t blame the Jaguars for not committing to one of rookies from the 2018 Draft. First overall pick Baker Mayfield was the only exceptional one, and taking him from the Cleveland Browns would have been impossible. Trading up for Darnold, Allen, and Rosen would have been difficult and quite possibly note worth the risk. That said, taking a slight but expensive upgrade from Bortles to Foles was not smart either.

Compared to Foles’s 28.84 mean QPR over the last two years, Bortles averaged a 28.23 in 16 games in 2017 and a poor 19.35 in a dozen 2018 starts, winning just 25% of those, for a grand 28-start average of 24.42 QPR with another wild standard deviation of 28.13. Foles is an upgrade over Bortles – statistics and tape will tell you that pretty quickly; however, how much of one is very debatable. One could easily say that Bortles had a stronger supporting cast in Philadelphia than what he will have in Jacksonville, and his Philadelphia mean QPR is not significantly higher than Bortles’s. If Foles can put up Bortles’s 2017 numbers in 2019, they could put together a good season, as he is only costing the Jaguars $12M in cap dollars this year; however, that changes significantly in 2020. If Jacksonville fails to win the Big Game in 2019, they will go into 2020 paying Foles over $22M in cap dollars without a Championship to show for it. That number increases to nearly $27M in 2021 and to that exact in 2022. They have a potential out in 2022 that will result in less than $7M of dead cap; however, cuts following 2019 season and before that out can result in dead cap numbers ranging from just under $34M in a pre-6/1/2020 cut to $6.25M in 2021 and 2022 if he were to be cut after 6/1/2021.

The Jaguars are married to Foles for the next three seasons, for better or for worse, and they would have probably been better off sticking with Bortles in 2019 and drafting his replacement early.

The Other Guys

Foles was not the only quarterback to move this offseason. We saw the Denver Broncos trade for Flacco, Washington trade for Case Keenum in the wake of Alex Smith’s career-threatening late-season injury, the Miami Dolphins sign Ryan Fitzpatrick, and the Tennessee Titans trade for Ryan Tannehill. All but Tannehill are expected to start in 2019, but if the past is any indicator of the future, Tannehill will see some opportunities to earn himself a chance with a new team in 2020. Comparing the quarterbacks’ productivity ratings with their new teams’ previous passers’, it is apparent that some teams did better than others, and today I just want to focus on the changes in Denver and Washington.

For the Broncos, they did find a slight upgrade from Keenum to Flacco. The former produced an average QPR of 23.68 in 16 games, with a win percentage of 0.375; however, Flacco’s average productivity rating hung in with the league average at 32.03 over nine games, winning four of them. Over the last three seasons, Keenum has averaged a 27.02 QPR, winning 53.85% of his games; similarly, Flacco has averaged nearly identically at a 27.36 QPR with a similar win percentage of 51.22%. So, statistically, 2018 Flacco was better than 2018 Keenum, and the trade should benefit the Broncos; however, how much of an upgrade is really up in the air, and the move will be a real head scratcher if the team does not invest heavily in a rookie quarterback when the NFL Draft arrives in Nashville at the end of April.

Speaking of Keenum, Washington had to find an upgrade over the group of backups that was trotted out on the field in 2018 after losing Smith. The trio of Colt McCoy, Mark Sanchez, and Josh Johnson had a combined average QPR of 0.26 in their six starts, with a 1-5 record (not counting any statistics where they came in mid-game). Keenum’s previously 23.68 mean is obviously superior to Washington’s backups, but looking into a game-by-game basis, we see that Keenum’s best nine starts are better than all of Johnson’s, and his best twelve performances are better than McCoy’s (all of Keenum’s games were better than Sanchez’s, which was the sixth-worst start of the entire 2018 season).

This begs the question: is Keenum an upgrade over Smith? The answer: probably not.

Smith won 60% of the games he started, which is better than Keenum’s 37.50%, but that could be due to a difference in supporting case, as Keenum’s mean QPR of 23.68 is insignificantly lower than Smith’s average of 24.45. It’s hard to say if Keenum would have been more or less productive in Washington than Smith, but since Keenum’s offensive weapons were superior and the Broncos’ defense was statistically very similar to Washington’s, I think we can assume that Washington would not have been as successful with the University of Houston product. Now they will find out what they can do with him, but any plan that does not involve drafting a quarterback early is not a good one.

Fitzpatrick is not necessarily an upgrade over Tannehill in Miami, as history has shown that Fitzpatrick’s good performances can be very fluky. While he did post a 41.24 mean QPR, his standard deviation of 42.01 is the largest since 2011, meaning that on any given game, he could easily post a productivity rating over 80 or below zero, which he did twice of each in his seven starts. His wild inconsistency will leave Miami with much to worry about on Sunday, especially if there is not a rookie on roster that can compete for the starting job.

With regard to Tannehill, his 20.15 mean QPR is better than Marcus Mariota’s 17.10, a career-low; however, as Paul Kuharsky reported on his website recently, Mariota was far more injured than previously conceived. I built the QPR to be viewed from a per-start basis rather than a per-snap basis because I feel that availability and health are major factors to a quarterback’s success, and some quarterbacks are better at avoiding injuries than others. That said, with Tannehill arguably being an injury-prone passer, I think it is fair to excuse Mariota’s poor productivity due to the gruesome injuries that he sustained, and we should also give Tannehill the benefit of the doubt. Mariota deserves another season to prove himself, which his contract is allots him, but should he not prove himself to be capable of being healthy and elite, the Titans should move on rather than re-sign long-term.

Wrapping it Up

It has been a while since I have written anything, so please forgive me if this was too long. I feel like my word count is absurdly long, yet I did not hit on everything that I would have liked to.

While NFL teams are learning and trying to win games while their quarterbacks are on their rookie contracts, the Patriots have proven that the best way to go to the Super Bowl and win championships is through an elite quarterback. Theoretically, teams could cycle through and re-draft quarterbacks every four years and perpetually build around the rookie contracts; however, there is so much more to the NFL than winning games, and the quarterback contributes to that. He is the favorite and least favorite guy on the team. He is the first jersey sold at the shoe store in the mall. He is a face and leader of the franchise, the city, and the NFL as a whole. He sells shoes, sports drinks, and home/auto insurance packages (note, do not go skimpy on your uninsured motorist insurance. Higher premiums are inexpensive and provide you with good coverage in the worst situations).

More than any of this, the NFL quarterback affects a football game more than any other guy on the field – more than the running back, the receivers, the offensive line, the defense, the referees, and debatably, the head coach. The NFL quarterback cannot lead a team to the playoffs on his own, but he can put the team on his back better than almost any other player can in any other sport. This clutch ability comes with experience, which is what the veteran quarterbacks have over the young bucks on their rookie deals.

The elite quarterback is so rare and necessary for continuous success in the NFL, and there is no price too steep for an elite quarterback. Some thought that Kansas City was wrong to trade up for Mahomes, and some (including myself) thought they were wrong to trade Smith and go all-in on the Texas Tech product. The great 20/20 hindsight shows that the price that Kansas City paid is miniscule compared to the success they saw in 2018 and what they are primed for over the next decade-plus, if Mahomes can play remotely as well as he did in his first season as a starter.

The franchise quarterback is overrated, but the elite one is not.